By continuing to browse this site you agree to us using cookies as described in About Cookies
Wiley Online Library is migrating to a new platform powered by Atypon, the leading provider of scholarly publishing platforms. The new Wiley Online Library will be migrated over the weekend of February 24 & 25 and will be live on February 26, 2018. For more information, please visit our migration page: http://www.wileyactual.com/WOLMigration/
Odds ratio (OR) and hazard ratio (HR) are listed with their 95% confidence intervals.
Correspondence: Wilmar M. Wiersinga, Department of Endocrinology & Metabolism, Academic Medical Center room F5-167, Meibergdreef 9, Amsterdam 1105AZ, The Netherlands. Tel.: +31 20 5666071;Fax: +31 20 6917682; E-mail: email@example.com
Current smoking in population surveys is associated with a slight dose-dependent fall of serum TSH, likely secondary to a rise of serum FT4 and FT3 induced by activation of the sympathetic nervous system; it is independent of iodine intake. In contrast, the slightly greater thyroid size in smokers is observed in iodine-deficient but not in iodine-sufficient areas and caused by competitive inhibition of thyroidal iodide uptake by thiocyanate. Smokers have an increased prevalence of nontoxic goitre and thyroid multinodularity, at least in iodine-deficient areas. Current smoking reduces dose dependently the risk of thyroid cancer, which is more pronounced for papillary than for follicular types; the risk in former smokers approaches that of never smokers. The lower TSH and lower body mass index in smokers might contribute to this reduced risk. Current smoking lowers the risk of developing thyroid peroxidase and thyroglobulin antibodies and subclinical and overt autoimmune hypothyroidism; the effect is dose dependent, but disappears within 3 years after quitting smoking. There is evidence from an animal model of experimental autoimmune thyroiditis that anti-inflammatory effects of nicotine are involved. In contrast, smoking is a dose-dependent risk factor for Graves’ hyperthyroidism and especially for Graves’ ophthalmopathy. Smoking is related to a higher recurrence rate of Graves’ hyperthyroidism, a higher risk on Graves’ ophthalmopathy after 131I therapy and a less favourable outcome of GO treatment with steroids or retrobulbar irradiation. The observed associations with smoking likely indicate causal relationships in view of consistent associations across studies, the presence of dose–response effects and disappearance of associations after cessation of smoking.
Smoking has significant effects on the thyroid in health and disease, which has been the subject of three reviews about a decade ago.[1-3] Since that time, a number of large population-based surveys and controlled studies have clarified some previously unsettled issues in the interaction between smoking and thyroid, but also identified new relationships like the inverse association between smoking and Hashimoto's thyroiditis. In contrast to the well-known risk of smoking for the development of Graves’ hyperthyroidism, smoking appears to protect against autoimmune hypothyroidism. The biologic mechanisms responsible for the diverse effects of smoking on the thyroid remain largely unknown. This may be not too surprising by realizing there are numerous (>4000) components in tobacco, including alkaloids (such as nicotine and anatabine), gases (like carbon monoxide) and carcinogens (e.g. polycyclic aromatic hydrocarbons, aldehydes, free radicals and solvents). But in this area, some progress is also being made. I will give an update on smoking and thyroid, focussing on studies published in the last 10 years. Particular attention will be given to the question whether observed associations indicate a cause-and-effect relationship.
Smoking and thyroid function in the adult population
In the past, the effect of smoking on thyroid function in healthy subjects – if any – remained inconclusive as study results were conflicting.[2, 3] Recent large population-based surveys provide convincing evidence that current smokers have slightly but significantly lower serum TSH levels than nonsmokers. The NHANES III survey in the USA demonstrates that active smokers (serum cotinine >15 ng/ml) have a TSH distribution shifted towards lower levels as compared to nonsmokers, whereas subjects with mild smoke exposure (cotinine 0·05–0·15 ng/ml) have TSH values in each stratum in between those of active smokers and nonsmokers. The 5th Tromso study in Norway also observed lower TSH levels in smokers than in nonsmokers (in males, 1·63 ± 0·88 vs 1·95 ± 1·04 mU/l and in females, 1·55 ± 0·86 vs 1·86 ± 1·01 mU/l, respectively), independent of the number of cigarettes smoked. Smokers as compared to nonsmokers in this study had higher serum FT4 (14·0 ± 2·2 vs 13·4 ± 2·4 pm) and higher FT3 (3·89 ± 0·79 vs 3·72 ± 0·67 pm). Similar results are reported in the Norwegian HUNT study: mean TSH was lower in current smokers (1·33 and 1·40 mU/l in women and men, respectively) compared with former smokers (1·61 and 1·61 mU/l) and never smokers (1·66 and 1·70 mU/l). In former smokers, TSH increased gradually with time since smoking cessation, reaching levels of never smokers after 5–10 years in women and after >18 years in men. Among current smokers, TSH was higher in subjects smoking <4 cigarettes daily as compared to smoking 8–12 cigarettes daily. Some but not all studies report lower TSH levels also in passive smokers.[7-9] One hour of moderate passive smoking increases T3 by 0·09 nm and FT4 by 2·32 pm but does not affect significantly TSH. Higher FT3 levels in smokers compared to nonsmokers are also observed in an Italian survey.
The effect of smoking on thyroid function may be confounded by body mass index (BMI) as TSH is positively associated with BMI,[12-14] and smokers may have lower BMI. The relationship between TSH and BMI has been reported as significant only in nonsmokers, as strong in current as in never smokers and stronger in smokers than in nonsmokers. However, any association of BMI with higher TSH and lower FT4[12, 16] is in the opposite direction of the association of smoking with lower TSH and higher FT4, rendering BMI as a confounder less likely. A Danish study compared thyroid function tests before and after mandatory iodization of salt in the year 2000: like before iodization, smokers had lower TSH and higher FT4 than nonsmokers. The mechanism by which smoking lowers serum TSH remains unknown. It is tempting to speculate that the fall of TSH is secondary to a rise of serum FT3 and FT4, as a short exposure to cigarette smoke for just 1 h increased serum FT3 and FT4 although not followed by a significant fall of serum TSH. Maybe the exposure period was too short to witness a fall of TSH. Smoking might thus increase thyroid hormone synthesis and release via TSH-independent pathways. One may speculate that activation of the sympathetic nervous system is involved via its innervation of the thyroid gland.[19, 20]
Current smoking lowers serum TSH by about 0·3 mU/l
the effect is dose dependent
the effect disappears slowly after cessation of smoking
the effect is not associated with ambient iodine intake
the effect is accompanied by a slight rise of serum FT3 and FT4
the effect might be mediated by activation of the sympathetic nervous system stimulating thyroid hormone synthesis and release
Smoking and hypothyroidism
A meta-analysis published in 2002 could not find an association between smoking and hypothyroidism. A decade later, several large population-based studies have provided strong evidence that current smoking protects against hypothyroidism. In NHANES III, the prevalence of TSH values >4·5 mU/l is lower in smokers than in nonsmokers (2·6% vs 5·5%; RR 0·5, 0·4–0·6), in a dose-dependent fashion. Current smoking was likewise inversely related to subclinical hypothyroidism in two community-based surveys in Korea and Iran.[21, 22] The largest study comes from the population-based HUNT study in Norway: as compared to never smokers, in current smokers, the odds ratios for subclinical and overt hypothyroidism were 0·54 (0·45–0·66) and 0·60 (0·38–0·95), respectively, in women; corresponding figures in men were 0·37 (0·26–0·52) and 0·51 (0·15–1·73). Odds ratios for former smokers were nonsignificant. The cause of hypothyroidism in these studies – although excluding people with previously known thyroid disease and thereby also cases of post-thyroidectomy and postradioiodine hypothyroidism – is not specified. Most probably, the cause was Hashimoto's hypothyroidism in the vast majority. Subsequent studies indeed demonstrated clearly that smoking protects against autoimmune hypothyroidism. Four studies report a lower prevalence of thyroid antibodies in current smokers. In the Amsterdam AITD cohort at baseline, there were 25% smokers in subjects with TPO-Ab and 38% in subjects without TPO-Ab (OR 0·69, 0·48–0·99). In NHANES III, TPO-Ab were present in 11% of smokers vs 18% in nonsmokers (OR 0·57, 0·48–0·67); there was a dose-dependent relationship. Similar results were obtained in Denmark (where the association was stronger for Tg-Ab than for TPO-Ab) and in Teheran.[22, 24] During follow-up of the Amsterdam AITD cohort, cessation of smoking by subjects who had no thyroid antibodies at baseline increased the risk of developing TPO-Ab and/or Tg-Ab. There was a significant trend towards more quitters in subjects who developed autoimmune hypothyroidism than in the group who remained euthyroid. In the prospective DanThyr study, 140 patients diagnosed as having autoimmune hypothyroidism were compared with 560 matched controls. Reported smoking habits were largely confirmed by cotinine levels in urine. Patients with autoimmune hypothyroidism had more often quitted smoking in the last 2 years before diagnosis than controls (16·4% vs 3·4%). The increased risk of autoimmune hypothyroidism upon quitting was, however, transient: odds ratios <1 years after stopping smoking, 1–2 years following cessation and 3–10 years after quitting are 7·36 (2·27–23·90), 6·34 (2·59–15·3) and 0·75 (0·30–1·87), respectively.
How could these protective effects of smoking be explained? Moderate alcohol consumption also protects against autoimmune hypothyroidism, but its effect is independent of smoking. It may well be that nicotine is involved. Nicotine acts via binding to nicotine receptors, which are expressed in the peripheral and central nervous system but also on immune cells like CD4+ T cells, dendritic cells and macrophages. Nicotine might increase Treg-mediated immune suppression of lymphocytes via its α7-nicotinic-acetylcholine receptor (nAChR). Nicotine reduces the severity of experimental autoimmune encephalomyelitis, shifting the autoimmune profile from pathogenic Th1 and Th17 responses to protective Th2 responses. The data suggest that nAChR agonists may be effective in the treatment of inflammatory disorders. The minor tobacco alkaloid anatabine has a structure similar to nicotine, but is – in contrast to nicotine – nonaddictive and nontoxic at therapeutic doses and has a longer 8-h half-life. Anatabine reduces the incidence and severity of experimental autoimmune thyroiditis induced by thyroglobulin: it reduced the incidence of Tg-Ab, improved recovery of serum T4 and suppressed nitric oxide synthase and cyclooxygenase-2 production by macrophages.
Smoking reduces the risk of Hashimoto's thyroiditis (diminishing the occurrence of TPO- and Tg-Ab and autoimmune hypothyroidism) by about 40%
the effect is dose dependent
the effect disappears a few years after cessation of smoking
the effect might be related to activation of nicotine receptors on immune cells, shifting the autoimmune profile away from Th1 and Th17 responses
Smoking and hyperthyroidism
A 2002 meta-analysis concluded that ever smoking was not associated with toxic nodular goitre (OR 1·27, 0·69–2·33). In contrast, smoking was recognized as a clear risk factor for Graves’ disease. For Graves’ hyperthyroidism, the OR was 3·30 (2·09–5·22) in current smokers compared with never smokers; ex-smokers had no significant excess risk (OR 1·41, 0·77–2·58). The risk is significantly higher in women than in men. The odds ratio associated with ever smoking for Graves’ ophthalmopathy is even higher (OR 4·40, 2·88–6·73). The effect is dose dependent: the relative risk of diplopia or proptosis is 1·8, 3·8 and 7·0 at 1–10, 11–20 and >20 cigarettes/day, respectively; the risk is no longer significant in ex-smokers. Even passive smoking carries a risk of Graves’ ophthalmopathy: the prevalence of thyroid eye disease among children with Graves’ hyperthyroidism is highest in countries with the highest prevalence of smoking among teenagers. In women participating in the HUNT study, the odds ratios for overt hyperthyroidism (as compared to never smokers) were 0·95 (0·44–2·06) in former smokers and 2·37 (1·34–4·20) in current smokers; corresponding figures for subclinical hyperthyroidism were 1·19 (0·66–2·16) and 1·83 (1·10–3·06). Odds ratios were still significant in subjects who had quitted smoking 0–3 years before diagnosis, but not if time elapsed since smoking cessation was ≥4 years. Cases in men were too small in number for a meaningful analysis. The figures likely refer to hyperthyroidism due to Graves’ disease because previously known thyroid disease had been an exclusion criterion. Similar results were obtained in women from the Nurses Health Study II. The adjusted hazard ratio of Graves’ hyperthyroidism was 1·93 (1·54–2·43) among current smokers and 1·27 (1·03–1·56) among past smokers, with a clear dose–response relationship. Hazard ratio in ever smokers increased gradually from 0·91 (0·66–1·26) at 1–5 pack-year to 1·97 (1·32–2·95) at 21–25 pack-year. The risk decreased slowly with the number of years since a past smoker had stopped smoking: compared with current smokers, HR was 0·83 (0·58–1·21) at 5 years since quitting, 0·58 (0·39–0·87) at >10–15 years and 0·52 (0·41–0·65) in never smokers. In this study, obesity was associated with a decreased risk of Graves’ hyperthyroidism, whereas physical activity and alcohol intake carried no risk. The latter finding is in contradiction with a recent finding that moderate alcohol consumption is associated with a considerable reduction in the risk of Graves’ hyperthyroidism, independent of smoking habits. Smoking behaviour furthermore adversely affects the outcome of Graves’ disease. Whereas serum concentrations of TBII do not differ between smokers and nonsmokers at diagnosis of Graves’ hyperthyroidism, smokers display a much slower reduction in TBII during treatment with antithyroid drugs.[36, 37] Smokers also have a higher recurrence rate of Graves’ hyperthyroidism.[38-41] In one particular study, recurrence risk in patients with negative TBII at discontinuation of antithyroid drugs was 57% in smokers and 18% in nonsmokers; if TBII were positive, recurrences occurred in 100% of smokers and 86% of nonsmokers. Smoking increases the risk of developing or worsening of eye changes fourfold after 131I therapy of Graves’ hyperthyroidism (RR 3·94, 1·41–11·02), and improvement in eye changes with steroids is about four times less (RR 0·23, 0·14–0·40).[42-44] The outcome of treating Graves’ ophthalmopathy with glucocorticoids or retrobulbar irradiation is less favourable in smokers relative to nonsmokers.[44, 45]
Cigarette smoking affects innate and adoptive immune mechanisms with both pro-inflammatory and immunosuppressive effects, but it is still obscure how smoking may enhance immune pathways like Th2 responses leading to Graves’ hyperthyroidism. Human orbital fibroblasts in culture increase hyaluronic acid production and adipogenesis when exposed to cigarette smoke extract in a dose-dependent manner.
Current smoking increases the risk of Graves’ hyperthyroidism about twofold and of Graves’ ophthalmopathy (GO) about threefold
the effect is dose dependent
the effect is more pronounced in women than in men
the effect disappears a few years after cessation of smoking
the effect might be related to stimulating Th2 responses Current smoking is associated with a higher recurrence rate of Graves’ hyperthyroidism, a higher risk on GO after 131I therapy and a less favourable outcome of GO treatment with steroids or retrobulbar irradiation
Smoking and thyroid size in the adult population
In the SU.VI.MAX cohort in borderline iodine-deficient France, thyroid volume was negatively correlated with urinary iodine and positively with urinary thiocyanate; mean thyroid volume was greater among current smokers (15·1 and 9·6 ml in men and women, respectively) and former smokers (13·7 and 9·2 ml, respectively) than in nonsmokers (12·1 and 8·6 ml, respectively). In contrast, cigarette smoking and alcohol consumption were not related to thyroid volume (9·9 ml in men and 6·6 ml in women) in the population of Barcelona where iodine intake is sufficient. Likewise, smoking was not related to thyroid volume in iodine-sufficient Istanbul, but positively related to thyroid volume and serum thyroglobulin in iodine-deficient Denmark (with strongest associations in areas with lowest iodine intake). The difference in thyroid volume between heavy smokers and nonsmokers in Denmark was reduced from 24% before to 12% after mandatory salt iodization in 2000. Data from Pomerania in Germany also show a declining impact of smoking on thyroid growth as iodine supply improves.
The effect of smoking on thyroid size is attributed to thiocyanate (a degradation product of cyanide in tobacco smoke), which is a competitive inhibitor of thyroidal iodide uptake: in a cell line stably transfected with the human sodium–iodide symporter, 50% inhibition of iodide uptake occurs at 19·3 μm thiocyanate. Serum thiocyanate concentrations in smokers are in the range of 80–120 μm.
Current smoking increases thyroid size by about 3 ml in men and 1 ml in women • the effect becomes smaller in former smokers • the effect is restricted to areas with iodine deficiency • the effect is attributed to thiocyanate, acting as a competitive inhibitor of thyroidal iodide uptake.
Smoking and nontoxic goitre
The goitrogenic effect of thiocyanate has been recognized for a long time. The prevalence of nontoxic goitre in Swedish women was higher in smokers than in ex-smokers and nonsmokers (8·4%, 6·4% and 6·4%, respectively, in women born in 1929 and 2·9%, 0·4% and 0·7% in women born in 1941). Smoking was positively associated with palpable goitre – OR 3·1 (1·6–5·8) – in Denmark, being strongest in areas with the most pronounced iodine deficiency. Goitre prevalence in former smokers was close to that of never smokers. The Danish study also showed an association between smoking and thyroid multinodularity – OR 1·9 (1·4–2·5) – but not with increased prevalence of solitary thyroid nodules. Prevalence of multinodularity was 16·5% in heavy smokers and 7·6% in nonsmokers, with intermediate figures for moderate smokers and ex-smokers, suggestive of a dose–response relationship. In a large cohort of parous Swedish women, smoking was also associated with nontoxic goitre and nodules – HR 1·26 (1·14–1·38) – but a dose–response effect was not observed. In a moderately iodine-deficient area of Turkey, heavy smoking was associated with increased prevalence of thyroid multinodularity and goitre as compared to moderate smoking, but in iodine-sufficient Istanbul, smoking did not affect significantly goitre development or nodule formation.
Current smoking carries a risk for nontoxic goitre and multinodularity • the effect is dose dependent • the effect disappears after cessation of smoking • the effect is most pronounced in iodine deficiency • the effect is attributed to the goitrogenic action of thiocyanate.
Smoking and differentiated thyroid carcinoma
A pooled analysis of 14 case–control studies published in 2003 demonstrated a reduced risk of thyroid cancer in ever smokers: odds ratios were 0·6 (0·6–0·7) in current smokers and 0·9 (0·8–1·1) in former smokers. There were significant trends of reduced risk with greater duration and frequency of smoking. The findings were consistent in males and females and in papillary and follicular carcinomas. Consumption of wine and beer (but not of coffee and tea) also decreased thyroid cancer risk, but the association was lost after adjustment for current smoking. Two prospective studies published between 2010 and 2012 reach similar conclusions. Among U.S. radiologic technologists followed from 1983 through 2006, current smoking (but not alcohol intake) was inversely associated with thyroid cancer risk in women (hazard ratio 0·54). In postmenopausal women enrolled in the Women's Health Initiative with a follow-up of 12·7 years, current smokers had reduced risk of papillary thyroid cancer (HR 0·34, 0·15–0·78) compared with never smokers; the risk in ever smokers was not altered. A clear dose–response relationship was not observed, probably because the number of current smokers among cases was small. Alcohol consumption did not appear to affect risk. A pooled analysis of five prospective U.S. studies found that compared with never smokers, current smokers had a reduced risk of thyroid cancer (HR 0·68, 0·55–0·85); the association was stronger among nondrinkers (HR 0·46, 0·29–0·74). The risk in former smokers was not reduced compared with never smokers. Greater smoking intensity, duration and pack-years were associated with further risk reductions. Associations were more pronounced for papillary than follicular cancers. Alcohol intake (≥7 drinks/week vs 0) was also inversely associated with thyroid cancer risk (HR 0·72, 0·58–0·90).
A case–control study conducted in New Caledonia is of special interest in view of the exceptionally high incidence rate of thyroid cancer in that area, particularly in Melanesian women in whom the prevalence of obesity is very high. Thyroid cancer was negatively associated with tobacco smoking and alcohol drinking, but no dose–response relationship was observed. However, in Melanesian women of 50 years and older, a strong positive association was found between thyroid cancer and body mass index (BMI): for a BMI ≥35 kg/m2, the odds ratio was 5·5 (1·5–20·3) compared with normal weight women, with a clear dose–response trend. Obesity as a risk factor for thyroid cancer was also evident in the U.S. radiologic technologists study (HR 1·74, 1·03–2·94) for BMI ≥35·0 vs BMI 18·5–24·9. The pooled analysis of five prospective U.S. studies also found strong evidence that obesity is an independent risk factor for thyroid cancer in both men and women and both papillary and follicular carcinomas. Compared with normal weight (BMI 18·5–24·9), hazard ratios for overweight (BMI 25·0–29·9) and for obesity (BMI ≥30) were 1·20 (1·04–1·38) and 1·53 (1·31–1·79), respectively. The risk was greater with increasing BMI: per 5 kg/m2 hazard ratios were 1·16 (1·08–1·24) in women and 1·21 (0·97–1·49) in men. These data might be relevant for understanding how smoking might decrease the risk of thyroid cancer. There is good evidence of a positive association between BMI and TSH: the higher BMI, the higher TSH. In NHANES 2007–2008, every 1-quartile increase in BMI increased serum TSH by 3·8% in men and by 4·0% in women. FT3 (but not FT4) also increased with every 1-quartile increase in BMI (1·0% in men and 1·3% in women). There is also convincing evidence that higher serum TSH is associated with higher risk of thyroid cancer. A recent meta-analysis found an odds ratio of 1·72 (1·42–2·07) per mU/l for TSH <0·1 mU/l and of 1·16 (1·12–1·21) per mU/l for TSH ≥1·0 mU/l. The model predicts a doubling of odds of thyroid cancer between TSH levels of 0·65 and 4·0 mU/l. Mean TSH levels in Korean patients with thyroid cancer were 0·3 mU/l lower than in controls (1·95 vs 1·62 mU/l); taking TSH 0·4–1·10 mU/l as reference, odds ratios of 2nd, 3rd and 4th quartiles (corresponding to TSH levels of 1·11–1·63, 1·64–2·30 and 2·31–4·8 mU/l) were 1·27, 1·55 and 2·21, respectively. One may thus speculate that the reduced risk of thyroid cancer in smokers (next to the anti-oestrogenic effect of smoking) is – at least to some degree – related to lower TSH levels and lower BMI associated with smoking. Numerous studies indicate that BMI is lower in cigarette smokers than in nonsmokers. However, heavy smokers tend to have greater body weight than do light smokers and nonsmokers. This likely reflects a clustering of risky behaviours in smokers, such as low physical activity, unhealthy diet and high alcohol intake. As so often, a number of possible confounders hinder to unravel the precise pathways by which the effects of smoking occur.
Current smoking reduces the risk of differentiated thyroid carcinoma by about 40% • the effect is dose dependent • the effect is more pronounced for papillary than follicular cancers • the effect may disappear after cessation of smoking • the effect might be related to some extent to lower TSH and lower body mass index in current smokers.
Smoking has distinct associations with thyroid function and size in healthy subjects. It has remarkable and contrasting associations with thyroid function in autoimmune thyroid disease (lower risk of Hashimoto's disease and higher risk of Graves’ disease) and with thyroid size in nodular disease (lower risk of thyroid carcinoma and higher risk of nontoxic goitre and multinodularity) (Fig. 1). The observed associations likely indicate causal relationships in view of consistent associations across studies, the presence of a dose–response relationship and disappearance of the associations after cessation of smoking. Which mechanisms mediate the many effects of smoking remains largely obscure. Probably, they differ between the various effects. The divergent effects of smoking on the expression of autoimmune thyroid disease are intriguing and reminiscent on the contrasting effects of smoking on inflammatory bowel disease: protective against ulcerative colitis (OR 0·41, 0·34–0·48) but risky for Crohn's disease (OR 1·61, 1·27–2·03). One is tempted to speculate that – given the genetic susceptibility of autoimmune thyroid disease – a subject may steer the course of his disease into the direction of either too much or too little thyroid hormone by adapting his smoking behaviour. The relationship between smoking and thyroid is frequently confounded by iodine intake and alcohol consumption, possibly also by other environmental factors like selenium intake. Further studies should take these factors into account. Gene–environment interactions have hardly been analysed. A notable exception is a report relating smoking and polymorphisms in detoxification genes and genes involved in DNA repair–apoptosis pathways to the risk of Graves’ disease and its outcome. The overriding risk of lung cancer and cardiovascular diseases associated with smoking warrants the recommendation to stop smoking in all patients with thyroid disease irrespective of their nature. Although a disappointingly low proportion of smokers is able to quit smoking, the physician should support such attempts. Recent guidelines containing such supportive measures are very helpful in this respect.[69, 70]